Leaderboard/hardware/m3-pro-18-core-gpu

M3 Pro (18-core GPU) — LLM benchmarks

Name: M3 Pro (18-core GPU) — community LLM benchmarks
Creator: llm-speed
License: https://www.apache.org/licenses/LICENSE-2.0
Keywords: M3 Pro (18-core GPU), LLM benchmark, tokens per second, decode tok/s, prefill, TTFT

2 workload results across 1 model.

Fastest known config on M3 Pro (18-core GPU)

286.5 decode tok/s

Qwen2.5-0.5B-Instruct-4bit via mlx (4bit) — see full run

Qwen2.5-0.5B-Instruct-4bit

Workload	Backend	Quant	decode tok/s	prefill tok/s	TTFT	Run
chat-short	mlx@0.31.3	4bit	286.5tok/s	656.2tok/s	200ms	r_akcbpx5vcqa
chat-short	mlx@0.31.3	4bit	282.6tok/s	668.8tok/s	196ms	r_bftqtkilvoe

Models measured on M3 Pro (18-core GPU)

Qwen2.5-0.5B-Instruct-4bit benchmarks